publish eve_rock_watcher.py

eve_rock_watcher.py

@@ -90,55 +90,57 @@ QUANTITY_RE = re.compile(r"quantit[yvſ]\s*[:.]?\s*([\d.,]{2,})\s*units?", re.I)
 UNITS_RE = re.compile(r"([\d.,]{3,})\s*units?\b", re.I)
 
 
-def _ocr_window_text(cp):
-    """OCR the EVE window with sparse-text mode (psm 11) — far better than uniform-block
-    (psm 6) at picking scattered UI text off a busy nebula background. Returns flat text."""
+def read_selected_quantity(cp):
+    """Return (units, ore) for the currently-selected rock from its floating info popup.
+    Two-pass: (1) locate the word 'Units' anywhere in the window via sparse OCR, then
+    (2) crop just that line, zoom + grayscale it, and OCR the digits cleanly — small text
+    over the nebula garbles in a full-window pass ('5b Units'), but reads fine zoomed in."""
     import pytesseract
+    from PIL import Image  # noqa: F401  (capture_window already returns a PIL image)
     tcmd = cp.get("ocr", "tesseract_cmd", fallback="").strip()
     if tcmd:
         pytesseract.pytesseract.tesseract_cmd = tcmd
     img = w.capture_window()
     if img is None:
-        return None
-    out = []
-    for psm in (11, 6):                                # sparse first, then block fallback
-        try:
-            out.append(pytesseract.image_to_string(img, config=f"--psm {psm}"))
-        except Exception:
-            pass
-    return "  ".join(out).replace("\n", " ")
-
-
-def read_selected_quantity(cp):
-    """Return (units, ore) for the currently-selected rock from its floating info popup,
-    found anywhere in the EVE window. (None, None) if no rock is selected/visible."""
-    flat = _ocr_window_text(cp)
-    if flat is None:
         return None, None
-    m = QUANTITY_RE.search(flat)
-    if not m:                                          # "Quantity" garbled? take any "N Units"
-        for cand in UNITS_RE.finditer(flat):
-            v = int(re.sub(r"\D", "", cand.group(1)) or 0)
-            if 100 <= v <= 5_000_000:                  # plausible asteroid remaining
-                m = cand
+    try:
+        d = pytesseract.image_to_data(img, config="--psm 11",
+                                      output_type=pytesseract.Output.DICT)
+    except Exception:
+        return None, None
+    n = len(d["text"])
+    cands = [i for i in range(n) if d["text"][i].strip().lower().startswith("unit")]
+    dbg = []
+    for i in cands:
+        uy, uh, ux = d["top"][i], d["height"][i], d["left"][i]
+        # crop the line just LEFT of "Units" — that's "Quantity <number>"
+        box = img.crop((max(0, ux - 320), max(0, uy - 8), ux + 6, uy + uh + 8))
+        z = box.convert("L").resize((box.width * 4, box.height * 4))
+        line = pytesseract.image_to_string(
+            z, config="--psm 7 -c tessedit_char_whitelist=0123456789,QuantiyUns ")
+        dbg.append(repr(line.strip()))
+        mm = re.search(r"([\d,]{3,})", line)
+        if not mm:
+            continue
+        v = int(re.sub(r"\D", "", mm.group(1)) or 0)
+        if not (100 <= v <= 5_000_000):               # plausible asteroid remaining
+            continue
+        # ore name = a token a little above the 'Units' line, same popup block
+        ore = "rock"
+        for j in range(n):
+            if d["text"][j].strip() and 0 < (uy - d["top"][j]) < 120 \
+                    and abs(d["left"][j] - (ux - 160)) < 260:
+                o = match_ore(d["text"][j].strip().lower())
+                if o:
+                    ore = o
                     break
-    if not m:
-        # leave a breadcrumb so we can see what OCR produced when it misses
+        return v, ore
     try:
         with open(os.path.join(w.HERE, "_rockocr.txt"), "w", encoding="utf-8") as fh:
-                fh.write(flat[:4000])
+            fh.write(f"units-candidates={len(cands)} crops={dbg}\n")
     except Exception:
         pass
     return None, None
-    digits = re.sub(r"\D", "", m.group(1))
-    if len(digits) < 2:
-        return None, None
-    # the rock's name sits just before its 'Quantity' line in the same popup block
-    # (e.g. "Omber  Distance ...  Quantity 47,765 Units") — match there, not the whole
-    # window (the hold's "Compressed Kernite" would otherwise win).
-    ore = match_ore(flat[max(0, m.start() - 180):m.start()].lower()) \
-        or match_ore(flat.lower()) or "rock"
-    return int(digits), ore
 
 
 def save_rock_region(cp, l, t, wd, ht, mode="survey"):